Supermicro and Nokia bring NVIDIA's AI-RAN compute to the cell site edge
As seen on the 24/7 Wall St. homepage on October 21, 2026.
Supermicro Brings NVIDIA Aerial RAN Computer Solutions to the Network Edge with Nokia to Enable Low-latency AI Services
Telecom carriers become a new GPU buyer class here: Nokia's radio software now rides on Supermicro boxes built around NVIDIA's Aerial RAN Computer, with the edge ARC-Pro node delivering up to 1.6 PFLOPS of FP4 inference at the cell site. The ARC MGX servers are generally available now, with the 2U ARC-Pro hitting general availability in the first half of 2027.
Continue ReadingShow less
Telecom carriers are now in the GPU business whether they planned to be or not. The three-way arrangement announced October 21, 2026 puts Nokia's radio access network software on Supermicro hardware built around NVIDIA's Aerial RAN Computer platform, letting operators run both connectivity and AI workloads on the same physical infrastructure rather than maintaining separate systems for each.
The edge ARC-Pro node delivers 1.6 PFLOPS of FP4 inference at the cell site itself, which makes low-latency applications like voice AI, vision AI, and autonomous network operations practical at the access layer. Pushing that compute out to the radio is the architectural shift the partnership is built around.
ARC MGX servers for distributed cell sites and centralized aggregation locations are generally available now, while the 2U ARC-Pro is targeted for customer testing later in 2026 and general availability in the first half of 2027. Carriers wanting full edge inference have roughly a year before deploying at scale.
Mobile operators represent an entirely new class of GPU buyer, largely separate from the hyperscaler and enterprise AI spending investors have followed closely. If carriers budget accelerated compute into RAN refresh cycles, the addressable market for both Supermicro systems and NVIDIA silicon expands well beyond the data center footprint that has driven growth so far.